agb(agent-browser)
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/cecwap/agbWhat This Skill Does
The agb (agent-browser) skill is a powerful automation tool integrated into the OpenClaw AI ecosystem, designed to bridge the gap between AI reasoning and the web. It provides a robust interface to control a Chromium-based browser, enabling the AI to navigate websites, interact with complex UI components, fill out forms, perform data extraction, and capture visual information. By utilizing accessibility trees and interactive element mapping (refs), the skill allows the AI to perceive the web much like a human would, mapping UI elements to temporary identifiers (e.g., @e1) to ensure precise interactions. This capability transforms a standard LLM into a sophisticated browser agent capable of complex task completion.
Installation
To integrate this skill into your OpenClaw environment, execute the following command in your terminal:
clawhub install openclaw/skills/skills/cecwap/agb
This will provision the necessary agent-browser components and the Chromium runtime. Please ensure your environment allows process management to utilize the browser efficiently.
Use Cases
- Web Automation Testing: Automate repetitive end-to-end testing of web applications, including login flows, navigation validation, and form submission checks.
- Automated Research: Automatically visit multiple URLs, extract key data points, and compile findings into a structured report.
- Dynamic Form Completion: Fill out lengthy web forms or surveys that require data mapping from local documents.
- Visual Monitoring: Capture screenshots of specific web elements or full pages for visual regression testing or tracking status changes over time.
- Data Scraping: Interact with modern, JavaScript-heavy websites to collect information that traditional static crawlers cannot access.
Example Prompts
- "Open https://example.com/login, fill in my username and password from the context, and take a screenshot after logging in."
- "Go to the company documentation page, find the search bar, type 'API authentication', and list all the links found in the results section."
- "Navigate to the registration portal, click the sign-up button, and extract the text from the error message that appears if I leave the form empty."
Tips & Limitations
- Session Management: Always use the
--sessionflag for parallel tasks to prevent state contamination. - DOM Instability: Remember that web pages are dynamic. Always call
snapshot -iafter any major navigation or action to ensure your@refsare current. - Resource Management: Chromium is resource-intensive. Be diligent about using the
closecommand and periodically runningpkill -f chromiumto reclaim system memory and prevent zombie processes. - Persistence: Use the
--profileflag if you need the agent to remain logged in across multiple sessions. Default execution is headless, which is faster but hides the UI; toggle this if you need to debug visual rendering issues.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-cecwap-agb": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: network-access, data-collection, code-execution